Surface water flooding risk to urban communities: Analysis of vulnerability, hazard and exposure

Landscape and Urban Planning 103 (2011) 185–197

Contents lists available at ScienceDirect

Landscape and Urban Planning journal homepage: www.elsevier.com/locate/landurbplan

Surface water flooding risk to urban communities: Analysis of vulnerability, hazard and exposure ∗ ´ Aleksandra Kazmierczak , Gina Cavan School of Environment and Development, University of Manchester, Oxford Road, Manchester M13 9PL, UK

a r t i c l e

i n f o

Article history: Received 9 February 2011 Received in revised form 13 July 2011 Accepted 25 July 2011 Available online 8 September 2011 Keywords: Surface water flooding Social vulnerability Land use Green space Risk assessment

a b s t r a c t Developing appropriate responses to address and prevent surface water flooding requires an analysis of interactions between elements of a risk framework encompassing hazard, vulnerability and exposure. This paper explores the spatial distribution of surface water flooding, the vulnerability of communities to flooding, and the characteristics of physical environment and land use that affect people’s exposure to flooding, particularly concerning green cover in Greater Manchester, UK. A set of 26 indicators known to influence the vulnerability of people to flooding was identified for spatial census units in Greater Manchester, and simplified with Principal Component Analysis to four underlying factors relating to material situation, diversity of communities and high proportion of children or elderly in the population. This was followed by an analysis of the presence and spatial distribution of surface water flooding areas, land use types, green cover and housing. Finally, the spatial associations between hazard, vulnerability and exposure were analysed. The results indicate that some of the most vulnerable people in Greater Manchester, namely the culturally diverse and materially deprived communities, are at high risk of flooding due to a convergence of factors related to socio-economic characteristics of the population, spatial distribution of the hazard, and the land use and housing types present in the area. Some adaptation responses tailored to the characteristics of the community and the environment they live in are discussed. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Weather and climate are inextricably linked to many aspects of human life. Extreme weather events in recent decades have emphasised the significance of hazards associated with climate and weather, such as high temperatures or flooding following intense precipitation (Beniston & Stephenson, 2004). Climate change is likely to result in increased incidences of extreme weather events (Smith et al., 2009), and it is these, rather than gradual change over long time-periods, that cause the most significant risk to humans and natural systems (EEA, JRC, & WHO, 2008; Meehl et al., 2000). Adaptation to the projected increasingly frequent occurrence of extreme weather events is particularly important in cities, where the high density of populations combined with urbanised land cover, can exacerbate climate change risks to people (Beniston & Stephenson, 2004; Lindley, Handley, Theuray, Peet, & Mcevoy, 2006). Preparing for an uncertain future can be approached through improving adaptation to current climate variability and extreme weather events (Burton, 1997). In order to provide appropriate

∗ Corresponding author. Tel.: +44 0161 2756832; fax: +44 0161 2756893. ´ E-mail addresses: [email protected] (A. Kazmierczak), [email protected] (G. Cavan). 0169-2046/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.landurbplan.2011.07.008

adaptation responses, it is necessary to understand the nature of risk and its spatial distribution. This paper, based on the case study of Greater Manchester, UK, aims to assess the risk of surface water flooding for urban communities, by investigating the spatial associations between the vulnerability of people, the occurrence of surface water flooding, and the physical characteristics of the natural and built environment.

1.1. Vulnerability, hazard and exposure as elements of risk The risks associated with climate and weather can be understood as an interaction of hazard, exposure, and vulnerability, forming a ‘risk triangle’ (Crichton, 1999, 2007). Specifically in relation to flood risk, a source – pathways – receptors model is being used (DEFRA & EA, 2006). These models are combined in Fig. 1 in relation to the risk of surface water flooding to people in urban areas. Climate hazard, or ‘source’, in the risk triangle framework relates to extreme weather events, such as intense rainfall causing surface water flooding. Vulnerability refers to the intrinsic characteristics of the hazards’ receptors (which can be people, infrastructure, economic activities, or other), and defines the extent to which these receptors are susceptible to harm from, or unable to cope with, hazards. The term ‘exposure’ can be defined as the nature and degree to which a receptor (the urban communities in this

186

A. Ka´zmierczak, G. Cavan / Landscape and Urban Planning 103 (2011) 185–197

Hazard (source): Surface water flooding

Vulnerability of

RISK

receptors : Urban communities

Exposure (pathway): Urban environment Fig. 1. Flood risk triangle (adapted from Crichton (1999) and DEFRA and EA (2006)).

study) is exposed to climate or weather hazards (Parry, Canziani, Palutikof, van der Linden, & Hanson, 2007). Thus, exposure, closely related to the concept of a flooding ‘pathway’ (DEFRA & EA, 2006), refers to the geographical location of a receptor, as well as the characteristics of the specific location that can exacerbate or reduce the magnitude of a hazard’s impact. According to this framework, for risks to be realised, the receptors and hazard need to coincide spatially. Further, the magnitude of risk depends on the level of vulnerability of the receptors, the nature of the hazard, and the physical characteristics of the environment defining the exposure (Lindley et al., 2006). 1.2. Hazard: surface water flooding Surface water flooding describes the combined flooding in urban areas during heavy rainfall. As such, it includes pluvial flooding (that results from rainfall-generated overland flow and ponding before the runoff enters any watercourse, drainage system or sewer, or cannot enter it because the network is full to capacity), sewer flooding, flooding from small open-channel and culverted urban watercourses, and overland flows from groundwater springs (Falconer et al., 2009). Surface water flooding is predominantly caused by short duration intense rainfall, occurring locally. Therefore, such floods are difficult to forecast, warn against and prepare for (Falconer et al., 2009; Golding, 2009). The UK Office of Science and Technology (Evans et al., 2004) estimate that 80,000 urban properties in the UK are currently at risk from surface water flooding, yielding average annual damages of £270 million, and that these numbers are likely to increase in the future as a consequence of climate change. Thus, the widespread character of this hazard suggests that it is important to analyse the risks of surface water flooding to people in urban areas. 1.3. Exposure: land use and housing The management of surface water flooding is hindered by the characteristics of urban drainage. The presence of large sealed surfaces in urban areas (such as buildings, roads, and car parks) raises the volume of surface water runoff (Ripl, 1995; Sanders & Phillipson, 2003). Conversely, green spaces reduce runoff volumes and rates by facilitating the infiltration of water into the ground, and through evapotranspiration of water back into the air (Ripl, 1995). Gill, Handley, Ennos, and Pauleit (2007) modelled surface water runoff from different types of land use and found that runoff increases with the proportion of built-up areas. For example, on sandy soils, for an 18 mm precipitation event in North West England, low-density residential areas (66% vegetated cover) were characterised by 32% runoff, compared with 74% runoff in town centres (20% vegetated cover). Therefore, it is apparent that land use, and the proportion of evapotranspiring surfaces in particular, has a significant influence on surface water behaviour, and therefore the exposure of the area to flooding. Further, whilst the focus of this study is on the risks to people, it is important to note that specific land uses and types of infrastructure can have a negative impact on the functioning of urban areas if they are affected by flooding.

Flooding of critical infrastructure, such as transport routes, electricity distribution substations, waste management facilities, water or sewage treatment works, can be both an inconvenience and a threat to urban residents. The presence of facilities such as hospitals or schools in areas at risk of flooding may put them at risk, but they can also function as places for shelter or as temporary accommodation in an emergency situation, if they are protected from flooding. Another element of exposure in the context of flooding is housing type. Houses with the lowest floor at or below ground level are more exposed than dwellings located on higher floors, and occupants and their belongings may be more significantly affected (Thieken, Muller, Kreibich, & Merz, 2005). Solid masonry buildings can withstand flooding without suffering major structural damage, whilst lightweight constructions may be more easily damaged (Sanders & Phillipson, 2003). Housing type also affects the flood protection measures that could be provided. Detached houses are relatively easy to protect from shallow flooding at the propertylevel with the use of door guards, air brick covers and waterproof skirts, whilst in high-density terraced housing, water can seep under flooring and through walls between adjacent properties (Bowker, 2002). Thus, only community-level flood protection measures (structural defences; landscaping) would be effective and financially feasible for terraced housing (Johnson & Priest, 2008). In this paper, the housing and land use types (with a particular focus on green spaces and presence of vegetation) are investigated as components of exposure to surface water flooding. 1.4. Vulnerability of people and communities to surface water flooding Knowledge of hazards aids the understanding of the physical aspects of disasters, but it may also result in a perception that extreme weather events are the disasters, indiscriminately affecting all residents of the location where it occurs (McEntire, 2005). However, some people are more susceptible to harm than others due to their different capacities to deal with hazards (Clark et al., 1998). Consequently, alongside the knowledge of hazards, awareness of the vulnerabilities to disasters is increasingly seen as a crucial factor in the reduction of disaster risk (McEntire, 2005; UNISDR, 2005). Vulnerability of people to flooding is sometimes understood simply as their ability to respond to a flood by being able to physically withstand the flood water’s velocity and depth (DEFRA & EA, 2006). However, in a broader sense, vulnerability of people to flooding is a function of the characteristics of people and households, which influence the following four types of issues: - Access to information. This has an impact on an individual’s awareness of living in an area at risk, on knowing what to do in the event of surface water flooding and understanding what help is available in the aftermath of a flood. Those without strong social networks and unfamiliar with their area (for example, as a result of short residence in the area or renting) may have less access to information (Cutter, Boruff, & Shirley, 2003). Inability to understand the information due to illiteracy (Cutter et al., 2003), no knowledge of the official language (McGeehin & Mirabelli, 2001), age or mental health, also increases vulnerability. - Ability to prepare for flooding. Those on low incomes may not be able to invest in flood insurance (Tapsell, Penning-Rowsell, Tunstall, & Wilson, 2002) or in property-level flood protection measures (Clark et al., 1998). Those physically less able, including the elderly, disabled or those suffering from a chronic illness (Clark et al., 1998) may find it difficult to secure their belongings from flooding once warnings are issued or in the early stages of a flood.

A. Ka´zmierczak, G. Cavan / Landscape and Urban Planning 103 (2011) 185–197

187

Fig. 2. Urban (over 66% built-up areas), suburban (34–66% built-up areas) and peri-urban (up to 33% built-up areas) areas in Greater Manchester. Based on Gill et al. (2007). Base map is ©Crown Copyright/database right (2009). An Ordnance Survey/EDINA supplied service.

- Ability to respond to flooding. The capacity of people to act in the event of flooding can be negatively affected by their young or old age (Clark et al., 1998; Cutter et al., 2003; Fernandez, Byard, Lin, Benson, & Barbera, 2002), pre-existing health problems (Rygel, O’Sullivan, & Yarnal, 2006), whether they care for others (e.g. lone parents) (Cutter et al., 2003; Tapsell et al., 2002), or if they own a car and can leave the area (Clark et al., 1998). Isolated and housebound people (especially the elderly) may wait longer for help when service providers cannot reach them due to impassable roads affected by the flood event (Fernandez et al., 2002). Similarly, those in need of additional support and resources (infirm, institutionalized, transient, or homeless) are disproportionately affected during disasters (Cutter et al., 2003). In areas of high density housing and where households are overcrowded, the emergency services responding to an event may be overstretched (Cutter et al., 2003). In the case of some ethnic minorities, cultural differences may hamper support from emergency services due to misunderstandings and problems with communication (Cutter et al., 2003). - Ability to recover. The populations in need of additional support and resources are often overlooked during recovery due to their relative invisibility in communities (Cutter et al., 2003). Those on lower incomes, carers, single parents, or pensioners may find it difficult to find the resources, energy and mental strength to start again (Clark et al., 1998; Tapsell et al., 2002). Children and the elderly are more susceptible to health-related impacts of floods (injuries, water pollution and other). Pre-existing problems with physical and mental health have been found to significantly affect the ability of people to recover after flooding (Tapsell et al., 2002). Also, older people and children have been found to suffer considerable psychological trauma following flood events (Fernandez et al., 2002; Rygel et al., 2006; Tapsell et al., 2002). Whilst spatial mapping of the vulnerability of people to extreme weather events has been the subject of studies in North America (Chakraborty, Tobin, & Montz, 2005; Clark et al., 1998; Cutter, Mitchell, & Scott, 2003; Cutter et al., 2000; Rygel et al., 2006;

Wu, Yarnal, & Fisher, 2002), fewer such studies have been undertaken to date in the UK context (Haynes, Haynes, & Pender, 2008; Lindley et al., 2006; Tapsell et al., 2002). The UK Cabinet Office (2008) emphasised that spatial representation of vulnerability with the use of GIS can be particularly useful for targeting emergency responses. DEFRA and EA (2006) offer an example of a methodology for assessment of the flood risk of people. However, this guidance only focused on the danger of people being swept over or drowned in a flood. Thus, there is a need for studies exploring the broader issues of vulnerability and their spatial relation to hazard and exposure. Consequently, this paper aims to spatially analyse the vulnerability of communities to surface water flooding – considering their ability to prepare for, respond to and recover after flooding – in Greater Manchester, UK. It then analyses the associations between the spatial distribution of vulnerability, exposure and hazard in the context of surface water flooding, in order to investigate the risk to communities. 2. Methods 2.1. Study area The conurbation of Greater Manchester in the North West of England (53◦ 30 N; 2◦ 15 W) covers 1276 km2 . It is located on a river basin flanked by the Pennine hills in the north and east and stretching to lowland areas to the south and west (Gill et al., 2007; Ravetz, 2000). Greater Manchester is a home to over 2.5 million people and it comprises ten districts: the cities of Manchester and Salford and eight metropolitan boroughs. The conurbation includes urban areas (town centres, retail and industrial areas), suburban housing areas, as well as peri-urban fringe (Fig. 2). Greater Manchester evolved from free standing industrial towns, during the industrial revolution associated with the cotton industry in the 19th century (Barlow, 1995). The conurbation continued to grow until 1960s until, following the global economic turn, it went through the process of post-industrial change, including significant job losses, rises in unemployment and poverty,

188

A. Ka´zmierczak, G. Cavan / Landscape and Urban Planning 103 (2011) 185–197

Table 1 Principal component (PC) loadings for the indicators of vulnerability. The most significant loadings for each variable are in bold. Aspect of vulnerability: 1 – access to information; 2 – ability to prepare; 3 – ability to respond; 4 – ability to recover. Variables

Aspects of vulnerability

% households with at least one person with a limiting long term illness % people deprived in terms of their incomea % people deprived in terms of their employmenta % households with no car % people with no or level 1 qualifications % of unemployed people among economically active 16–74 years old % households rented from social landlords % households with no adults in employment and dependent children % lone parent households with dependent children % people with limiting long-term illness % people whose health was not good Number of people per hectare % overcrowded households % households rented from private landlords % people born outside the Great Britain % ethnic minorities % children under 4 years old in the population % people 5–17 years old in the population % households with dependent children 5–18 years old % households with dependent children under 4 years old % single person households (non-pensioner) % people aged over 75 in the population % people aged 65–75 in the population % single pensioner households % all pensioner households % households with no adults in employment (no children) % residents living in residential care Initial eigenvalues Percentage of variance explained (%)

2, 3, 4

PC1: poverty 0.807

0.017

0.192

0.434

2, 4 2, 4

0.864 0.947

0.387 0.078

0.211 0.003

0.024 0.084

3 1, 2, 3, 4 2, 4

0.891 0.807 0.838

0.360 −0.112 0.368

−0.156 0.368 0.062

0.072 0.131 −0.116

1, 2, 3, 4 2, 3, 4

0.921 0.799

0.096 0.307

−0.030 0.368

0.046 −0.175

2, 3, 4

0.821

0.015

0.292

−0.236

2, 3, 4 2, 3, 4 3 2, 3, 4 1, 2, 3, 4 1, 3 1, 3 3, 4 3, 4 3

0.797 0.869 0.088 0.377 0.145 0.019 0.067 0.384 0.317 0.001

−0.056 0.027 0.523 0.814 0.657 0.936 0.918 0.317 0.029 0.034

−0.124 −0.108 0.105 −0.144 −0.499 0.114 0.202 0.653 0.828 0.898

0.523 0.372 −0.072 −0.110 −0.206 −0.052 −0.048 −0.300 −0.203 −0.360

3

0.192

0.335

0.773

−0.352

1, 3 1, 3, 4 1, 3, 4 1, 3, 4 1, 3, 4 2, 4

0.509 −0.007 0.150 0.390 −0.463 0.581

0.380 −0.055 −0.141 −0.039 −0.377 0.075

−0.630 −0.285 −0.215 −0.233 0.093 −0.342

−0.248 0.756 0.889 0.788 0.629 0.656

2, 3, 4

0.010 10.49 38.87

0.335 6.09 22.55

−0.335 3.73 13.82

0.138 1.63 6.03

PC2: diversity

PC3: children

PC4: old age

Source of data: ONS (2002). a CLG (2008).

and an increase in the magnitude of social problems (Barlow, 1995). In recent decades, urban regeneration in Greater Manchester has made the conurbation the largest cluster for finance, law, media, research and higher education in England outside of London. However, due to the out-movement of wealthier residents to the suburbs and peri-urban areas, the conurbation is still characterised by great income inequalities (Ravetz, 2000). Moreover, throughout its history, Greater Manchester has been a multi-ethnic centre for many groups, more recently including South Asian, Caribbean and Chinese (Ravetz, 2000). Therefore, many aspects of the socio-economic vulnerability of individuals and households in this post-industrial conurbation can be transferable to other British and European cities. An analysis of media reports for the period 1961–2009 in Greater Manchester found that flooding was a dominant climate impact within the conurbation, and the principal cause of damage to property and infrastructure (Smith & Lawson, 2011). Also, there seems to be an increase in the number of flood events over the decades. Whilst between 1961 and 1970 six pluvial flood events were recorded, their number increased to eight in the period 1981–1990 and 14 between 2001 and 2009 (Smith & Lawson, 2011). Whilst the press is inconsistent in recording weather, and recording of weather events by other organisations (e.g. local authorities) is generally not comprehensive, these findings seem to reflect the effects of recent changes in weather and climate. Also, future climate change projections suggest increased frequency and magnitude of precipitation events. Palmer and Räisänen (2002) estimate that the probability of total winter precipitation exceeding two standard

deviations above normal will increase by a factor of five over the North West of England. Specifically for Greater Manchester, analysis of the UK Climate Projections 2009 indicates that by the 2050s, under the high emissions scenario, changes in average winter precipitation across Greater Manchester are unlikely to be less than 0.3% (10th percentile) and unlikely to be more than 36% (90th percentile), and changes in rainfall on the wettest day in winter are unlikely to be less than 1.4% (10th percentile) and unlikely to be more than 38% (90th percentile) (Cavan, 2010). Therefore, the problems associated with heavy rainfall causing surface water flooding in Greater Manchester are likely to intensify.

2.2. Analysis of vulnerability Table 1 lists the 26 indicators describing vulnerability of people and communities to surface water flooding, reflecting their access to information and their ability to prepare for, respond to, and recover after flooding (see Section 1.4). The indicators were obtained from Census 2001 (ONS, 2002), and the Indices of Multiple Deprivation 2007 (CLG, 2008) for the Lower Super Output Area (LSOA) level. LSOAs are compact areas of homogenous socioeconomic characteristics constrained by the boundaries of the electoral wards used by the Office of National Statistics to report small area statistics across England and Wales. LSOAs contain on average a population of around 1500 people (circa 600 households), and a minimum population of 1000 residents (400 households) (ONS, 2008). There are 1646 LSOAs in Greater Manchester.

A. Ka´zmierczak, G. Cavan / Landscape and Urban Planning 103 (2011) 185–197

In order to avoid double-counting of inter-correlated variables and to ensure against the redundancy of data, the dataset was reduced with the use of Principal Component Analysis (PCA). PCA is a vector space transformation, which helps to identify patterns in high-dimensional data and reveals the underlying factors (principal components, further referred to as PCs) that best describe variations in the data through identification and clustering of variables that measure the same theme. Varimax rotation was used in order to maximize the variance of loadings, thus aiding the classification of variables to PCs (Rygel et al., 2006). This data reduction method results in zero correlations between the principal components. As a result of the PCA, each of the LSOAs in Greater Manchester is characterised by the values of a limited number of PCs, rather than by values of 26 indicators. Low values of the emergent PCs indicate low vulnerability, and high values represent high vulnerability. 2.3. Analysis of associations between vulnerability, hazard and exposure The geospatial map of Areas Susceptible to Surface Water Flooding 2009 (produced by JBA Consulting and licensed by the Environment Agency for emergency planning purposes) was used as the best spatial representation of areas at risk of surface water flooding available at the time of the study. The map is based on the Digital Terrain Model (5 m × 5 m resolution) and was produced by simulating a 1 in 200 year event for a 6.5 h duration rainfall event, as such rainfall would overwhelm even the most modern drainage system and so any impact from the drainage system can be ignored and is not accounted for in the model (GeoStore, no date). The map presents four categories indicating areas of increasing natural susceptibility to surface water flooding, based upon flooded depths of up to 0.1 m (no flooding); 0.1–0.3 m; 0.3–1.0 m and over 1.0 m (GeoStore, no date). In this paper, all areas where the depths of 0.1 m are exceeded are considered susceptible to surface water flooding, as even shallow floods can cause significant damage to housing and incur costs, thus making it difficult for people to recover after flooding (ABI, no date; Penning-Rowsell & Green, 2000). The areas where the flood depth may exceed 1 m represent the most extreme surface water flooding scenario, and are referred to as highly susceptible to flooding. Flood depths greater than 1 m may damage the structure of buildings (Soetanto & Proverbs, 2004), and thus displace vulnerable populations. Further, depending on the velocity of the flood water, depths between 0.5 m and 1.5 m may make it impossible for a person to stand (DEFRA & EA, 2006), thus presenting a direct risk to lives. Whilst, in the model, overland flow routes and areas where surface water ponds were identified using advanced 2D hydraulic modelling techniques (GeoStore, no date), the model does not include the velocity of flood waters. This could be seen as a limitation if the study focused exclusively on the ability of people to withstand the flood waters (DEFRA & EA, 2006), rather than the broader understanding of vulnerability, where it is likely to play a lesser role. In addition, Soetanto and Proverbs (2004) observe that the velocity of flood waters in the UK is not very high in general, and is not considered very destructive to properties by building surveyors (compared to its depth and contents), thus suggesting that it does not affect the material aspects of the ability to recover after flooding. The Greater Manchester Urban Morphology Type (UMT) dataset was used in order to gain an understanding of land use types present, as well as to analyse the proportion of green space in LSOAs. UMTs are homogenous urban land use types (Table 4), which were identified by the Centre for Urban and Regional Ecology at the University of Manchester within the Adaptation Strategies for Climate Change in the Urban Environment (ASCCUE) project (2003–2006). Gill et al. (2007) used aerial photograph interpretation of random points to estimate the mean percentage of nine land cover types

189

(trees, shrubs, mown grassland, rough grassland, cultivated, water, bare soil, building, and other impervious surfaces) in each UMT category. Here, the total area of green space in each UMT was calculated by summing the proportion of trees, shrubs, mown grassland, rough grassland, cultivated land, and water. The percentage of the territory of the individual LSOAs at risk of flooding was calculated by overlaying the Areas Susceptible to Surface Water Flooding 2009 with LSOAs in ESRI ArcGIS. Land use characteristics for each LSOA were investigated by overlaying this information with the UMT dataset, with a particular focus on green spaces, and with Ordnance Survey Mastermap (garden data). Data on the proportions of housing types in LSOAs (semi- and detached houses, terraced houses, houses with the lowest floor at or below ground level) was obtained from the Census 2001 (ONS, 2001). The percentage of houses in poor condition at the LSOA level was obtained from the Indices of Multiple Deprivation 2007 (CLG, 2008). The association between the vulnerability principal component scores, the proportion of areas at risk of flooding, and different types of housing, land use, green space and gardens in LSOAs was analysed using Spearman’s rank correlation. A one-way ANOVA was used to compare the mean proportion of areas of exposure and hazard in LSOAs between the most vulnerable areas (top quantile of the principle component scores). This was done in order to learn whether the areas with high vulnerability, exposure and hazard coincide, and which types of communities are at highest risk of surface water flooding. The overall risks for communities, characterised by selected aspects of vulnerability, were spatially mapped.

3. Results 3.1. Vulnerability of people and communities The Principal Component Analysis identified four principal components, which explained over 81% of variance in the data (Table 1). The components were named: poverty (PC1), which grouped variables associated with unemployment, material deprivation and poor health; diversity (PC2), which grouped variables associated with non-British origin, high density populations and private renting; children (PC3); and old age (PC4). One of the 26 original variables – the percentage of people living in residential care homes – did not load heavily on any of the principal components. This may be because a very low percentage of people live in care homes in the majority of the LSOAs. It is important to note that some of the variables are associated with more than principal component. For example, the percentage of children aged 0–4, whilst contributing the most strongly to PC3: children, was also associated with PC1: poverty and PC2: diversity (loadings exceeding 0.3). Similarly, the health related indicators (% people with limiting long-term illness, % households with at least one person with a limiting long term illness; % people whose health was not good), whilst loading the most heavily on the first factor associated with poverty and poor health, also contributed to PC4: old age. Thus, whilst the names of the principal components refer to the main reasons for vulnerability, careful interpretation of the results in Table 1 also reveals that these also include secondary aspects. The LSOAs with high scores of the poverty component in Greater Manchester are in general spatially concentrated around town centres, but there seem to also be pockets of deprivation and poor health in peri-urban areas (Fig. 3; compare Fig. 2). The LSOAs with high diversity component scores are even more visibly associated with the urban centres across Greater Manchester. Conversely, areas characterised by high children component scores (PC3) are more concentrated in sub- and peri-urban parts of the

190

A. Ka´zmierczak, G. Cavan / Landscape and Urban Planning 103 (2011) 185–197

Fig. 3. Principal component scores for Lower Super Output Areas in Greater Manchester (classified into quantiles). Base map is ©Crown Copyright/database right (2009). An Ordnance Survey/EDINA supplied service.

conurbation. The LSOAs with high scores of the old age component (PC4) are scattered throughout the suburban areas of Greater Manchester. 3.2. Hazard: distribution of surface water flooding Over 14.2% of the Greater Manchester area is susceptible to surface water flooding, and 2.2% is highly susceptible. The mapping of surface water flooding against the territorial units suggests that the hazard is widespread (Fig. 4): only five of 1646 LSOAs are not affected by surface water flooding and over 78% of LSOAs in Greater Manchester include areas highly susceptible to flooding. The areas

exposed to surface water flooding constitute up to 72.3% of individual LSOAs’ territory (mean = 16.3; SD = 10.4), whilst the percentage of individual LSOAs highly susceptible to flooding range between 0 and 20.2% (mean = 1.9; SD = 2.7). The spatial distribution of the percentage of LSOAs affected by surface water flooding across Greater Manchester (Fig. 4) indicates that the highest proportion of areas highly susceptible to flooding are located to the north of the conurbation, whilst the south-west and areas around urban centres are at lower risk of flooding. This varied distribution suggests that the levels of risk of surface water flooding are determined by factors associated with topography and land use.

Fig. 4. Proportion of LSOAs susceptible to surface water flooding (classified using natural breaks). Base map is ©Crown Copyright/database right (2009). An Ordnance Survey/EDINA supplied service.

A. Ka´zmierczak, G. Cavan / Landscape and Urban Planning 103 (2011) 185–197

191

Fig. 5. Spatial distribution of the total green space and gardens in Greater Manchester (classified using natural breaks). Base map is ©Crown Copyright/database right (2009). An Ordnance Survey/EDINA supplied service.

Table 2 Housing type and condition in Greater Manchester (% of total housing in LSOAs; N = 1646). Type of houses

Minimum

Maximum

Mean

Std. deviation

% semi- and detached houses % terraced houses % houses with lowest floor at/below ground level % houses in poor condition

1.0 0.0 28.3 8.0

99.6 95.4 100.0 57.0

53.4 32.0 88.2 27.0

27.1 22.1 11.5 8.1

3.3. Exposure of communities to surface water flooding The predominant land use category in Greater Manchester is green and open space, a large percentage of which is farmland (Table 4; see Fig. 2). Residential areas, predominantly of medium density development, form another important land use, hence the high proportion of suburban areas in Fig. 2. The total green space cover in Greater Manchester, calculated as the percentage of green and open space in all UMTs, is 71.7%, ranging 20.1–94.1% (mean = 58.7; SD = 14.6) in individual LSOAs. The highest percentage areas of green space are associated with peripheral areas of Greater Manchester (Fig. 5). Specifically, private gardens constitute 17.7% of the total Greater Manchester area, and their proportion in LSOAs ranges between 0.2 and 63.7% (mean = 25.9; SD = 14.2). The highest proportion of gardens is present in the suburban parts of the conurbation, in particular to the south of the conurbation (Fig. 5; compare Fig. 2). Over half of the housing in Greater Manchester comprises semidetached and detached housing, and terraced housing accounts for a third of the housing stock (Table 2). Consequently, the vast majority of housing across the conurbation has the lowest floor at or below ground level, which is significant when considering flood risk. Just over a quarter of housing in Greater Manchester is described as in poor condition.

correlation is not significant with regard to high susceptibility to flooding. Further, in the case of PC2, the more vulnerable that the communities are, the lower the proportion of areas highly susceptible to flooding is within LSOAs. The children component scores (PC3) are negatively correlated with the proportion of LSOAs at risk of flooding (including high susceptibility); and, in the case of the old age component scores (PC4), this holds true for all areas susceptible to flooding. Thus, the spatial distribution of communities characterised by poverty and poor health and high diversity coincides spatially with the areas at risk of surface water flooding. 3.4.2. Vulnerability of communities and exposure Analysis of Spearman’s rank correlations between the scores of principal components and the percentage of different types of land use in LSOAs (Table 4) reveals that an increase in vulnerability associated with poverty (PC1) is correlated with a decreasing proportion of different types of green space (except for formal open space, which increases with vulnerability), as well as with decreasing peri-urban characteristics (such as the presence of farmland). With increased scores of the poverty component, the proportion of high density residential areas, roads, retail areas, and town centres in LSOAs increased. In addition, higher scores of the poverty component are associated with greater proportions of derelict and underused land, as well as with existing manufacturing areas. Similarly, an increase in scores of the diversity factor (PC2) is associated

3.4. Associations between vulnerability, hazard and exposure This section analyses the associations between the elements of the risk triangle (see Fig. 1): vulnerability of communities, the hazard of surface water flooding and the exposure factors relating to the land use and housing characteristics in order to identify the character of risks present. 3.4.1. Vulnerability of communities and hazard The poverty and diversity component scores (PC1 and PC2) are positively associated with the proportion of LSOAs susceptible to surface water flooding (Table 3). However, in the case of PC1, the

Table 3 Spearman’s rank correlations between the level of vulnerability and the proportion of LSOAs exposed to surface water flooding in Greater Manchester (N = 1646). Principal components

All areas susceptible to flooding

High susceptibility

PC1: poverty PC2: diversity PC3: children PC4: old age

0.056* 0.139*** −0.099*** −0.111***

Ns −0.234*** −0.059* Ns

Ns – not significant. * Significant at 0.10 level. *** Significant at 0.01 level.

192

A. Ka´zmierczak, G. Cavan / Landscape and Urban Planning 103 (2011) 185–197

Table 4 Spearman’s rank correlations between the percentage of different land use types in LSOA, the percentage of areas exposed to surface water flooding in LSOAs and the scores of principal components (N = 1646). Land use category (% in GM)

Land use type

Principal components

% of LSOA susceptible to surface water flooding

PC1: poverty

PC2: diversity

PC3: children

PC4: old age

All areas susceptible

High susceptibility

Industry and business (8.8%)

Manufacturing Offices Distribution and storage Mineral works/quarries Retail Town centre

0.198*** Ns Ns Ns 0.103*** 0.134***

−0.050* 0.101*** −0.083** −0.123*** Ns 0.306***

−0.071** −0.113*** −0.074** Ns −0.092*** −0.183***

−0.087*** −0.057* Ns Ns Ns Ns

0.127*** 0.083** 0.092*** Ns 0.089*** 0.070**

0.195*** Ns 0.074** 0.114*** 0.088*** Ns

Green/open space (55.5%)

Improved farmland Unimproved farmland Disused/derelict land Remnant countryside Woodland Formal recreation Formal open space Informal open space Allotments Rivers and canals

−0.248*** −0.091*** 0.167*** −0.065** −0.182*** −0.054* 0.066** Ns Ns Ns

−0.532*** −0.195*** −0.247*** −0.186*** −0.218*** Ns 0.104*** −0.057* Ns −0.108***

Ns Ns Ns Ns Ns Ns −0.087*** Ns Ns −0.107***

−0.058* −0.111*** −0.124*** Ns Ns 0.063** 0.102*** Ns Ns −0.084**

−0.239*** −0.152*** 0.073** Ns −0.171*** 0.049* Ns 0.074** 0.055* 0.209***

0.181*** 0.078** 0.215*** 0.148*** 0.228*** 0.057* Ns 0.106*** Ns 0.274***

Critical infrastructure (1.0%)

Energy production and distribution Water storage/treatment Refuse disposal

Ns −0.077** Ns

−0.089*** −0.193*** Ns

Ns Ns Ns

Ns Ns −0.058**

0.053* 0.048* Ns

0.091*** 0.173*** 0.058*

Residential areas (30.0%)

High density Medium density Low density

0.214*** −0.051* −0.372***

0.413*** Ns −0.081**

−0.158*** 0.181*** −0.053*

−0.062* 0.114*** 0.077**

0.052* Ns −0.074**

−0.057* −0.230*** Ns

Social infrastructure (2.3%)

Cemeteries/crematoria Schools Hospitals

Ns Ns Ns

Ns −0.057* Ns

−0.053* 0.063* −0.064*

Ns 0.085** 0.049*

Ns Ns Ns

Ns Ns Ns

Transport infrastructure (2.4%)

Major roads Airports Rail

0.077* Ns Ns

Ns Ns Ns

−0.080** Ns −0.067**

Ns Ns Ns

0.136*** Ns 0.162***

0.110*** Ns 0.194***

Ns – not significant. GM – Greater Manchester. * Significant at 0.10 level. ** Significant at 0.05 level. *** Significant at 0.01 level.

with the increase in urban land use characteristics (proportion of high density housing, town centre and offices) and a decrease in farmland. Increasing scores of the children component (PC3) are associated with the increase in the proportion of medium- and low-density housing and with a decreasing proportion of “urban” land uses such as retail, offices, manufacturing or distribution, and transport infrastructure. The communities characterised by a high score of the old age component (PC4) are associated with lower amounts of both peri-urban land use types (farmland), as well as high density housing, manufacturing, derelict land and office areas. An increase in the score of the old age component is also associated with an increase in the proportion of medium- and low density residential areas in LSOA, as well as formal recreation and green spaces, schools and hospitals. Analysis of the correlations between the component scores and the amount of green space in LSOAs (Table 5) reveals that increasing scores of components of poverty and diversity are associated with a reduction in green space. In addition, in areas characterised by high scores of the poverty component, the presence of gardens is lower than in less vulnerable areas. Conversely, in areas inhabited by communities characterised by a high score of the diversity component, the percentage of gardens is higher than in areas characterised by lower scores. There are more gardens in areas with higher proportions of families with dependent children and elderly populations (high scores of PC3 and PC4). Thus, whilst the urban character of the environment in areas inhabited by the communities characterised by high vulnerability due to poverty or diversity can increase their

exposure to surface water flooding, the exposure of elderly populations and populations with a high proportion of children may be reduced due to a high proportion of green space. Increasing scores of the poverty, diversity and old age components are associated with decreasing proportions of houses with the lowest floor at ground level (Table 5). Conversely, communities vulnerable due to a high proportion of children in the population (high scores of PC3) are more likely to live in areas where houses have the lowest floor at ground level, comprising mostly of semi- and detached housing. The scores of poverty and diversity factors (PC1 and PC2), are positively correlated with the percentage of terraced housing and houses in poor condition in LSOAs. In contrast, the children component is negatively associated with the proportion of houses in poor condition. Thus, whilst the exposure of communities with a high proportion of children is increased by the ground-level character of housing, the actual damage may be reduced by the good quality of housing.

3.4.3. Hazard and exposure Significant positive associations are present between the percentage of LSOA susceptible to surface water flooding and the percentage of areas covered by the land use types associated with industry and business, critical infrastructure and transport-related land use types (Table 4). In the case of critical infrastructure, these correlations are stronger regarding high susceptibility to flooding. The proportion of LSOA susceptible to flooding shows an inverse relationship with the proportion of residential areas; this is

A. Ka´zmierczak, G. Cavan / Landscape and Urban Planning 103 (2011) 185–197

193

Table 5 Values of Spearman’s rank correlation coefficient between the principal component scores and presence of green space in LSOAs (N = 1646).

% total green space % gardens % houses with lowest floor at/below ground level % semi- and detached houses % terraced houses % houses in poor condition

PC1: poverty

PC2: diversity

PC3: children

PC4: old age

% area at risk of flooding

% area highly susceptible to flooding

−0.269*** −0.269*** −0.347*** −0.617*** 0.535*** 0.382***

−0.559*** 0.199*** −0.553*** −0.449*** 0.209*** 0.543***

0.056* 0.231*** 0.457*** 0.337*** −0.080** −0.372***

Ns 0.113*** −0.110*** 0.055* −0.130*** Ns

−0.152*** Ns −0.131*** −0.076** Ns 0.069**

0.237*** −0.365*** 0.064** Ns 0.048* Ns

Ns – not significant. * Significant at 0.10 level. ** Significant at 0.05 level. *** Significant at 0.01 level. Table 6 Comparison of variables of exposure (mean proportion of types of housing and green spaces) and hazard (mean proportion of areas susceptible to flooding) between the most vulnerable areas (top quantile of each factor). Factors of exposure and hazard

% houses with lowest floor at/below ground level % semi- and detached houses % terraced houses % houses in poor condition % green space % gardens % area susceptible to flooding % area highly susceptible to flooding

Highest vulnerability (top 20% of PC scores)

F

PC1

PC2

PC3

PC4

83.50 35.41 38.72 28.02 54.04 22.85 16.62 1.89

77.56 32.73 40.35 34.12 47.73 25.97 19.29 1.34

91.46 56.97 33.76 25.30 56.64 29.29 15.79 1.67

86.38 55.78 25.14 26.91 58.64 28.91 14.48 1.62

75.079*** 85.312*** 30.951*** 77.067*** 41.800*** 14.158*** 11.349*** Ns

Ns – not significant. *** Significant at 0.01 level.

particularly visible in the case of medium density residential areas and areas highly susceptible to flooding. LSOAs characterised by a high percentage of houses with the lowest floor at or below ground level are located in areas highly susceptible to flooding rather than in areas with a general risk of flooding (Table 5). Also, although correlations are weak, results suggest that larger proportions of housing in poor quality and terraced housing are at greater risk of flooding than other housing types. This may suggest that the hazard and exposure factors coincide spatially, potentially exacerbating the risks. Table 5 suggests that the areas with a large proportion of land susceptible to surface water flooding tended to have less green space; however, the larger the proportion of a LSOA that is highly susceptible to flooding, the more green space and fewer gardens are present. The associations between the proportion of area at risk of

flooding and proportion of green space are explained for individual green space types in Table 4: whilst the percentage of woodland and farmland in LSOAs decreases with the increasing proportion for areas highly susceptible to surface water flooding; the association is positive for derelict and disused land, formal recreation sites and informal open space. 3.4.4. Comparing hazard and exposure among the most vulnerable communities A one-way ANOVA comparing the exposure and hazard characteristics between the LSOAs occupied by communities classified as the most vulnerable to surface flooding due to the highest scores of the four principal components supports the results of the Spearman’s rank correlation (Table 6). The communities vulnerable due to high percentages of elderly (PC4) and children (PC3) are located

Fig. 6. Spatial distribution of the risk of surface water flooding to urban communities (classified into quantiles). Base map is ©Crown Copyright/database right (2009). An Ordnance Survey/EDINA supplied service.

194

A. Ka´zmierczak, G. Cavan / Landscape and Urban Planning 103 (2011) 185–197

in LSOAs characterised by the lowest proportionate area susceptible to flooding; the highest proportion of green space and gardens and the lowest proportion of houses in poor condition. The only factor of exposure that situated them at a disadvantage (in flood risk terms) is the high proportion of detached housing, and high proportion of houses with the lowest floor at or below ground level. Conversely, LSOAs occupied by communities vulnerable due to poverty (PC1) and diversity (PC2) are characterised by a significantly larger proportion of area highly susceptible to flooding, combined with a higher proportion of houses in poor condition and lower proportion of green space (in particular PC2) and gardens (in particular PC1). 3.4.5. Risk of surface water flooding to communities The three elements of risk: vulnerability, hazard and exposure, were combined in order to demonstrate how the overall risk could be calculated and presented at the conurbation level. Fig. 6 provides example risk maps, based on the spatial coincidence of hazard, exposure and high levels of vulnerability of urban communities. Fig. 6 indicates that risk is highest where the largest proportions of area are highly susceptible to surface water flooding (see Fig. 4), high vulnerability of communities due to poverty and poor health or high cultural diversity (see Fig. 3), and high exposure (low proportions of green space (see Fig. 5), and high proportions of houses with the lowest floor at or below ground level) coexist. 4. Discussion and conclusions The research findings presented in this paper have revealed different facets of vulnerability of people and communities to surface water flooding through a Principal Component Analysis. This sets this research apart from the previous UK studies focused on vulnerability, which integrated social and financial factors into one index (e.g. Haynes et al., 2008; Lindley et al., 2006; Tapsell et al., 2002). The approach used here allows evaluation of not only where the most vulnerable people live, but also what makes them vulnerable. This can help local authorities, emergency services and other agencies, to provide information and assistance in preparation for, response to and recovery from flooding, appropriate to the community characteristics. Some aspects of vulnerability, namely poverty and diversity of communities, reflect the general socio-economic inequality of urban areas represented in spatial terms, and these two principal components together explained the highest proportion of the variance in the dataset (Table 1). However, this study has also emphasised that the percentage of children and elderly people in the population will also influence the ability of communities to cope with surface water flooding. Therefore, the vulnerability of populations in Greater Manchester to surface water flooding does not seem to be entirely synonymous with social exclusion or material deprivation, as previous studies in the UK suggested (see e.g. Haynes et al., 2008; Tapsell et al., 2002). The spatial distribution of different aspects of vulnerability varies across the conurbation. The poor and diverse communities tend to be associated with urban centres, and the communities characterised by higher proportions of wealthier families and the elderly are more linked to suburban and more peri-urban locations (Fig. 3). This is consistent with the associations between the level of vulnerability and the proportion of different land use types in LSOAs (Table 4); whilst high scores of components of poverty and diversity are associated with urban conditions (town centres, manufacturing areas, high density housing), the children component is clearly suburban in character (largely medium-density residential areas with little industry and business-oriented land use). This pattern neatly reflects the processes of socio-spatial diversification of

cities present in most of Western Europe and North America, but particularly visible in British cities, and associated with the continuing trend of wealthier populations moving out of inner-city areas to more attractive, suburban locations, and the subsequent void being filled by private renters and immigrants (Dorling & Rees, 2003; Healey, McNamara, Elson, & Doak, 1988). Further, the industrial history of Greater Manchester is visible in the relationships between the distribution of poor communities and the location of current manufacturing areas and derelict and underused land, which is frequently associated with previous industrial uses. The consistency between the distribution of vulnerable communities across Greater Manchester and these recognised urban processes suggests that similar patterns of vulnerability may be found in other cities. Although variables such as local climate, topography and institutional frameworks are also significant driving factors, the responses to surface water flooding in relation to these types of areas in Greater Manchester are likely to be transferable. Further, close associations between different aspects of vulnerability in particular geographic locations emphasise the potential role that spatial planning could perform in moderating exposure as factors increasing or decreasing flood risk. Analysis of the correlations between the vulnerability of communities and the proportion of areas at risk of surface water flooding suggests that communities vulnerable due to poverty and diversity are more exposed to the widespread risk of surface water flooding than wealthier and more homogenous communities (Tables 3 and 6). This suggests that the material loss and inconvenience associated with surface water flooding affect those with the least material means and the lowest ability to act or receive help in the event of disaster and in its aftermath (Tapsell et al., 2002). Thus, whilst this research has emphasised that vulnerability is a multi-faceted phenomenon, strong links remain between social inequality and climate change risk. In contrast, the negative relationship between the vulnerability of populations associated with a high proportion of elderly and dependent children shows that, whilst these populations can be considered vulnerable due to their socio-demographic characteristics, the low threat of surface water flooding means that they are at relatively lower risk in comparison to other vulnerable groups. The high proportion of built-up areas, associated with typically urban land uses (Table 4), may exacerbate the risk of flooding in the event of extreme precipitation (Gill et al., 2007; Sanders & Phillipson, 2003), and significant inequalities between different vulnerable groups were found in relation to the proportion of green space in LSOAs (Table 5). Whilst the communities vulnerable due to a high proportion of children or elderly in the population enjoy more green spaces and gardens in their vicinity, the areas vulnerable due to poor and diverse communities in general contain less green space (excluding gardens in the case of diverse communities). This may have repercussions not only for the risk of surface water flooding in these areas, but also for the general well-being of communities. Green spaces provide multiple benefits for urban communities, including: improvement in physical and mental health, contribution to building stronger communities, and improved image of the neighbourhood (Tzoulas et al., 2007). Therefore, this pattern of distribution of green space strongly suggests environmental injustice in Greater Manchester. In addition, only the areas identified as vulnerable due to a high proportion of elderly people seemed to be positively associated with the presence of schools and hospitals, which could be used as temporary shelters in the event of a flood. This may place the other vulnerable communities at a disadvantage in the event of a flood due to a lack of access to such facilities. However, more detailed assessment is needed in order to assess the distribution of infrastructure associated with social services and care in relation to flooding and vulnerability.

A. Ka´zmierczak, G. Cavan / Landscape and Urban Planning 103 (2011) 185–197

It is also notable that the proportion of land in LSOAs linked to economic functions, such as critical infrastructure and transport, is associated with the percentage of LSOAs at risk of flooding (Table 4). This highlights the risk of considerable economic losses for manufacturing, retail, distribution and storage, and businesses in town centres in the case of extreme precipitation and surface water flooding events. Further risks include that flooding of strategic transport routes can paralyse a city, and flooding of refuse disposal sites can result in contamination of water supplies, creating additional hazards to urban populations. The negative correlations between green space cover and the proportion of an area susceptible to flooding (Table 5) concur with previous studies (e.g. Gill et al., 2007), suggesting that the increasing amount of sealed surfaces in an area exacerbates the problem of flooding through increased runoff and reduced infiltration capacity. Positive correlations between the proportion of green space and the areas at high risk of flooding in LSOAs (Table 5) may be a result of some areas being known as prone to flooding and, therefore, considered unsuitable for development and maintained as open space. Further, the positive association between the proportion of areas at risk of surface flooding with formal recreation sites and informal open space (Table 4) indicates that these places may already be performing important functions as rain water collection, storage and infiltration areas, and should be afforded further protection from development for this reason. The role of private gardens in mediating the risk of flooding can be seen as particularly important, as they are negatively associated with the risk of flooding (Table 5). Gardens form a significant proportion of the green space in Greater Manchester and many other cities, and are the most widespread type of green space in the urbanised neighbourhoods of Greater Manchester (Gill et al., 2007). Gardens therefore have a great impact on the environmental performance of the conurbation. Approximately 40% of all the evapotranspiring surfaces in ‘urbanised’ Greater Manchester occur in residential areas, with medium-density residential areas accounting for the majority of such surfaces (Gill et al., 2007). The fact that the poorest communities have the fewest gardens (Table 6) again stresses the environmental injustice associated with unequal distribution and provision of green space across the conurbation and suggests that the capacity of the areas inhabited by these communities to deal with intense rainfall may be compromised. The risk of surface water flooding to communities vulnerable due to a high proportion of children in the population could be exacerbated by the high proportion of houses with the lowest floor at or below ground level in the most vulnerable areas (Table 6). However, as a lower proportion of houses tend to be in poor condition in these areas than elsewhere, the damage caused by flood events may be less than expected. Conversely, communities vulnerable due to poverty and their diverse character are more likely to live in areas with a high proportion of houses in poor condition. Various aspects of the vulnerability of people and communities to surface water flooding in Greater Manchester have been found to correspond with different levels of hazard and exposure to such events (Table 6). The communities vulnerable due to a high proportion of children or elderly in the population are found to be at significantly lower risk of flooding than other types of vulnerable communities, due to lesser flood hazard where they live, and the fact that their exposure to surface water flooding is buffered by a higher proportion of vegetation and better quality housing than in other vulnerable areas. In general, the communities vulnerable due to their diverse character are also living in areas most affected by surface water flooding. These communities are also the most exposed to flooding as a result of the low proportion of green space and the poor condition of housing in their area. The communities vulnerable due to poverty and poor health are characterised by only marginally lower occurrence of hazards and exposure. Therefore,

195

whilst the study has revealed different aspects of vulnerability, it can be concluded that the risk of surface water flooding (composed of the associations between vulnerability, exposure and hazard) is the highest for the communities in urban areas that are already socio-economically disadvantaged. 4.1. Developing responses tailored to exposure and vulnerability The vulnerability associated with a high percentage of children and elderly in the population coincided with a high proportion of houses exposed to flooding due to their lowest floor being located at or below ground level, and a high proportion of semi-detached and detached houses in the area. This, combined with the fact that the vulnerability of these communities is not related to material deprivation, suggests that the risk of surface water flooding in these areas could be lowered through the promotion of propertylevel flood protection measures, such as flood gates or air brick covers (Kazmierczak & Bichard, 2010). Conversely, in areas where poor or ethnically diverse, transient populations live in terraced houses and houses in poor condition, actions aimed at promoting the protection of individual properties against flooding may not succeed due to the tenure dominated by social and private renting, and the high costs associated with such solutions (Kazmierczak & Bichard, 2010). Also, the protection of individual houses in these areas may be counterproductive because water can seep through the walls between adjoining properties, and the options for floodproofing internal walls are limited (Bowker, 2002). Consequently, wider flood measures to protect entire streets and neighbourhoods should be promoted in such areas, and the vulnerability of people may be considerably reduced through the provision of relevant information about reducing risk of flooding to residents, in appropriate languages and format. Green spaces are considered to be an important measure for reducing the surface water runoff, thus limiting the impacts of extreme weather events on cities (Gill et al., 2007). Greening may include development and conservation of parks and woodlands, planting of moisture-retaining tree and shrub species, or construction of sustainable urban drainage systems, including ponds and reservoirs (Williams, Joynt, & Hopkins, 2010). The positive associations between the susceptibility to flooding and the proportion of informal open spaces and formal recreation sites in LSOAs (Table 4) suggest that these types of green spaces could potentially be used to provide the functions which help to manage surface water flooding. Where green space is absent, and the density of development does not allow for provision of new green areas, green roofs could provide an additional measure to reduce surface water flooding. Gill et al. (2007) found that adding green roofs in high density residential areas can reduce total runoff by up to 20%. Williams et al. (2010) recommend removal of non-porous driveways and restoration of green front gardens to reduce the risk of surface water flooding. This is particularly important in Greater Manchester as domestic gardens account for nearly one-fifth of the entire conurbation, and in many suburban areas they are the main type of green space (Gill et al., 2007). However, gardens are often lost as urban areas densify. For example, Pauleit, Ennos, and Golding (2005) observed a 5% loss in garden areas in Merseyside between 1975 and 2000 due to garden-grabbing by developers. The role of vegetation in surface water management recognised by Gill et al. (2007) emphasises the importance of “leafy suburbs” remaining as such, and calls for the introduction of anti-densification measures. The recent policy change in the UK removes gardens from the list of land use types classified as brownfield land, where development is encouraged (CLG, 2010); therefore, their further loss may be slowed down or stopped. The GIS analysis at the conurbation scale can successfully highlight areas that will require action in relation to climate change

196

A. Ka´zmierczak, G. Cavan / Landscape and Urban Planning 103 (2011) 185–197

related risks in the urban environment (Lindley et al., 2006). However, it needs to be emphasised that the results of the statistical analysis identified the principal reasons for vulnerability in a given area, rather than providing a detailed picture of the community characteristics. Thus, the use of principal components allows for strategic planning of adaptation actions across the conurbation, and a more detailed property-level analysis of vulnerability is needed to provide assistance to individual people and households. Similarly, more detailed flood modelling, incorporating velocity of flood water and debris content (see DEFRA & EA, 2006) for smaller areas, and also including the drainage system response under flooding scenarios less extreme than the 1 in 200 years event, which considered here due to the paucity of other data, would be advisable to estimate the magnitude of hazard more precisely. 4.2. Concluding remarks This research has helped to identify spatial patterns and associations between elements of hazard, vulnerability, exposure and risk across Greater Manchester, which could be indicative of a situation in other cities. Whilst different aspects of vulnerability have been identified, it was found that particularly for poor and diverse communities vulnerability coincides with the occurrence of the hazard of surface water flooding, and is exacerbated by exposure factors linked to land use characteristics and housing types. Surface water flooding already causes considerable damage in England (Evans et al., 2004) and Greater Manchester (Smith & Lawson, 2011), and continuing urbanisation combined with projected future climate changes will exacerbate the problem of surface water flooding (Cavan, 2010; Evans et al., 2004; Pitt, 2008), and therefore urgent adaptation responses are needed. These should take into account the social and demographic characteristics of populations, together with the land use and housing characteristics of the areas they occupy, in order to develop responses tailored to communities and considering both land use and housing improvements. Acknowledgements The research reported in this paper was carried out as part of the EcoCities research project at the University of Manchester, funded by charitable donations from Bruntwood Ltd. and The Oglesby Trust. The authors would like to acknowledge the Environment Agency for providing surface water flooding data, and Ordnance Survey for the use of spatial data, including MasterMap. Thanks go to Dr Jeremy Carter for his comments on the early version of the manuscript. References ABI. (no date). Flood resilient homes. What homeowners can do to reduce flood damage. London: Association of British Insurers. Available from: www.abi.org.uk/ Information/Consumers/General/15274.pdf Accessed 08.07.11. Barlow, M. (1995). Greater Manchester: Conurbation complexity and local government structure. Political Geography, 14, 379–400. Beniston, M., & Stephenson, D. B. (2004). Extreme climatic events and their evolution under changing climatic conditions. Global and Planetary Change, 44, 1–9. Bowker, P. (2002). Making properties more resistant to floods. Municipal Engineer, 151, 197–205. Burton, I. (1997). Vulnerability and adaptive response in the context of climate and climate change. Climatic Change, 26, 185–196. Cabinet Office. (2008). Identifying people who are vulnerable in a crisis. Guidance for emergency planners and responders. London: Civil Contingencies Secretariat, Cabinet Office. Cavan, G. (2010). Climate change projections for Greater Manchester. Manchester: University of Manchester. Available from: http://www.sed.manchester.ac.uk/ architecture/research/ecocities/library/documents/Climate change projections GM final.pdf Accessed 08.07.11. Chakraborty, J., Tobin, G., & Montz, B. (2005). Population evacuation: Assessing spatial variability in geophysical risk and social vulnerability to natural hazards. Natural Hazards Review, 6, 23–33.

Clark, G., Moser, S., Ratick, S., Dow, K., Meyer, W., Emani, S., et al. (1998). Assessing the vulnerability of coastal communities to extreme storms: The case of Revere, MA, USA. Mitigation and Adaptation Strategies for Global Change, 3, 59–82. CLG. (2008). The English indices of deprivation 2007. London: Communities and Local Government. CLG. (2010). Planning policy statement. 3. Housing. London: Communities and Local Government. Crichton, D. (1999). The risk triangle. In J. Ingleton (Ed.), Natural disaster management (pp. 102–103). London: Tudor Rose. Crichton, D. (2007). What can cities do to increase resilience? Philosophical Transactions of the Royal Society, 365, 2731–2739. Cutter, S. L., Boruff, B. J., & Shirley, W. L. (2003). Social vulnerability to environmental hazards. Social Science Quarterly, 84, 242–261. Cutter, S. L., Mitchell, J. T., & Scott, M. S. (2000). Revealing the vulnerability of people and places: A case study of Georgetown County, South Carolina. Annals of the Association of American Geographers, 90, 713–737. DEFRA, & EA. (2006). R&D outputs: Flood risks to people. Phase 2. FD2321/TR1 The flood risks to people methodology. London: Department for Environment Food and Rural Affairs and the Environment Agency. Dorling, D., & Rees, P. (2003). A nation still dividing: The British census and social polarisation 1971–2001. Environment and Planning A, 35, 1287–1313. EEA, JRC, & WHO. (2008). Impacts of Europe’s changing climate – 2008 indicator-based assessment. Copenhagen: European Environment Agency. Evans, E., Ashley, R., Hall, J., Penning-Rowsell, E., Saul, A., Sayers, P., et al. (2004). Foresight. Future flooding. Scientific summary: Volume 1 – Future risks and their drivers. London: Office of Science and Technology. Falconer, R. H., Cobby, D., Smyth, P., Astle, G., Dent, J., & Golding, B. (2009). Pluvial flooding: New approaches in flood warning, mapping and risk management. Journal of Flood Risk Management, 2, 198–208. Fernandez, L. S., Byard, D., Lin, C.-C., Benson, S., & Barbera, J. A. (2002). Frail elderly as disaster victims: Emergency management strategies. Prehospital and Disaster Medicine, 17, 67–74. GeoStore. (no date). JBA comprehensive flood map product. Available from: http:// www.geostore.com/geostore4/WebStore?xml=geostore4/xml/products JBAFLOOD.xml Accessed 24.11.10. Gill, S. E., Handley, J. F., Ennos, A. R., & Pauleit, S. (2007). Adapting cities for climate change; the role of green infrastructure. Built Environment, 33, 115–133. Golding, B. W. (2009). Long lead time flood warnings: Reality or fantasy? Meteorological Applications, 16, 3–12. Haynes, H., Haynes, R., & Pender, G. (2008). Integrating socio-economic analysis into decision-support methodology for flood risk management at the development scale (Scotland). Water Environment Journal, 22, 117–124. Healey, P., McNamara, P., Elson, M., & Doak, A. (1988). Land use planning and the mediation of urban change. The British planning system in practice. Cambridge: Cambridge University Press. Johnson, C., & Priest, S. (2008). Flood risk management in England: A changing landscape of risk responsibility? International Journal of Water Resources Development, 24(4), 513–525. Kazmierczak, A., & Bichard, E. (2010). Investigating homeowners’ interest in property-level flood protection. International Journal of Disaster Resilience in the Built Environment, 1, 157–172. Lindley, S. J., Handley, J. F., Theuray, N., Peet, E., & Mcevoy, D. (2006). Adaptation strategies for climate change in the urban environment: Assessing climate change related risk in UK urban areas. Journal of Risk Research, 9, 543–568. McEntire, E. (2005). Why vulnerability matters. Exploring the merit of an inclusive disaster reduction concept. Disaster Prevention and Management, 14, 206–222. McGeehin, M. A., & Mirabelli, M. (2001). The potential impacts of climate variability and change on temperature-related morbidity and mortality in the United States. Environmental Health Perspectives, 109, 185–189. Meehl, G. A., Karl, T., Easterling, D. R., Changnon, S., Pielke, R., Jr., Changnon, D., et al. (2000). Introduction to trends in extreme weather and climate events: Observations, socioeconomic impacts, terrestrial ecological impacts, and model projections. Bulletin of the American Meteorological Society, 81, 413–416. ONS. (2002). Census data 2001 dataset. London: Office of National Statistics. ONS. (2008). Names and codes for Super Output Area geography. London: Office for National Statistics. Available from: http://www.ons.gov.uk/about-statistics/ geography/products/geog-products-area/names-codes/soa/index.html Accessed 10.07.11. Palmer, T. N., & Räisänen, J. (2002). Quantifying the risk of extreme seasonal precipitation events in a changing climate. Nature, 415, 512–514. Parry, M. L., Canziani, O. F., Palutikof, J. P., van der Linden, P. J., & Hanson, C. E. (2007). Climate change 2007: Working Group II: Impacts, adaptation and vulnerability contribution of Working Group II to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, 2007. Cambridge, UK/New York, USA: Cambridge University Press. Pauleit, S., Ennos, R., & Golding, Y. (2005). Modelling the environmental impacts of urban land use and land cover change – A study in Merseyside, UK. Landscape and Urban Planning, 71, 295–310. Penning-Rowsell, E. C., & Green, C. (2000). New insights into the appraisal of flood-alleviation benefits: (1) Flood damage and flood loss information. Water Environment Journal, 14, 347–353. Pitt, M. (2008). Learning lessons from the 2007 floods: An independent review by Sir Michael Pitt. London: Cabinet Office. Ravetz, J. (2000). City region 2020. London: TCPA and Earthscan. Ripl, W. (1995). Management of water cycle and energy flow for ecosystem control: The energy-transport-reaction (ETR) model. Ecological Modelling, 78, 61–76.

A. Ka´zmierczak, G. Cavan / Landscape and Urban Planning 103 (2011) 185–197 Rygel, L., O’Sullivan, D., & Yarnal, B. (2006). A method for constructing a social vulnerability index: An application to hurricane storm surges in a developed country. Mitigation and Adaptation Strategies for Global Change, 11, 741– 764. Sanders, C. H., & Phillipson, M. C. (2003). UK adaptation strategy and technical measures: The impacts of climate change on buildings. Building Research and Information, 31, 210–221. Smith, C., & Lawson, N. (2011). Identifying extreme event climate thresholds for Greater Manchester, UK: Examining the past to prepare for the future. Meteorological Applications, 18 doi:10.1002/met.252 Smith, J. B., Schneider, S. H., Oppenheimer, M., Yohe, G. W., Hare, W., Mastrandrea, M. D., et al. (2009). Assessing dangerous climate change through an update of the Intergovernmental Panel on Climate Change (IPCC) reasons for concern. Proceedings of the National Academy of Sciences of the United States of America, 106, 4133–4137. Soetanto, R., & Proverbs, D. G. (2004). Impact of flood characteristics on damage caused to UK domestic properties: The perceptions of building surveyors. Structural Survey, 22, 95–104.

197

Tapsell, S. M., Penning-Rowsell, E. C., Tunstall, S. M., & Wilson, T. L. (2002). Vulnerability to flooding: Health and social dimensions. Philosophical Transactions of the Royal Society of London A, 360, 1511–1525. Thieken, A. H., Muller, M., Kreibich, H., & Merz, B. (2005). Flood damage and influencing factors: New insights from the August 2002 flood in Germany. Water Resources Research, 41 doi:10.1029/2005WR004177 Tzoulas, K., Korpela, K., Venn, S., Yli-Pelkonen, V., Kazmierczak, A., Niemelä, J., et al. (2007). Promoting ecosystem and human health in urban areas using Green Infrastructure: A literature review. Landscape and Urban Planning, 81, 167–178. UNISDR. (2005). Hyogo framework for action 2005–2015. Building the resilience of nations and communities to disasters. United Nations International Strategy for Disaster Reduction. Available from: http://www.unisdr.org/wcdr/ Accessed 24.11.10. Williams, K., Joynt, J. L. R., & Hopkins, D. (2010). Adapting to climate change in the compact city: The suburban challenge. Built Environment, 36, 105–115. Wu, S.-Y., Yarnal, B., & Fisher, A. (2002). Vulnerability of coastal communities to sealevel rise: A case study of Cape May County, New Jersey, USA. Climate Research, 22, 255–270.